Sections:
- Blank Workflow
- Component Library
- Adding Components
- Visualizer Components
- Saving the Workflow
- Section Summary
Creating a Workflow
In this section we will step through the creation of a simple workflow from components in the component library. The workflow we will create contains only two components, the ``Aimed Collection Reader'' and the visualization component ``Annotation Viewer''. This workflow will allow us to see graphically the gold standard annotations produced by the collection reader.
Blank Workflow
The first thing we need is a blank workflow in the Workflow Viewer. To achieve this simply click the ``New Workflow'' button in the workflow menu.
In a blank workflow, the workflow viewer will contain two empty boxes, one labeled ``Collection Reader'' and the other ``Analysis Engines and CAS Consumers''. It is into these boxes that we will be adding components to build our workflow.
Component Library
The components we want to add can be found in the Component Library. Have a look for the ``Aimed Collection Reader'' which is located under the English : Collection Readers directory of the library. Once you find this component click to select it.
Selecting the component will cause it's details to be displayed in the description panel below the component list. Here, the components name and a short description of what it does are given. Additionally there is an iconic representation of what type of component it is and lists of any inputs and outputs. In the case of the ``Aimed Collection Reader'' its type is Collection Reader, there are no inputs, but there are four output types listed.
Looking at the rest of the component library you will see that U-Compare comes with a large selection of components of various types. Later sections of this guide will cover how to import additional components to the library and how to combine existing components together to create new composite components.
Adding Components
We are now going to add the ``Aimed Collection Reader'' to the workflow. To do this simply click and drag its entry from the component library into the box labelled ``Collection Reader'' in the workflow viewer. When you have done this the workflow viewer should appear as follows:
The AIMED reader has now been added to the collection reader slot of the workflow. The representation of the collection reader in the viewer is similar to that in the Component Library description panel, but also displays details on configuration parameters specific to this instance of the component. There are also two buttons in the upper right hand corner for editing and removing the component from the workflow. We will come back to these later, but for now will continue with the components default settings.
The component we have just added reads in files from the AIMED corpus and also provides gold standard annotations for several features. What we are going to do now is add a component that will allow us to visualize these annotations. This component is the ``Annotation Viewer'' and it is located in the Visualizers directory of the component library. Once you have located it now drag and drop it into the box in the workflow viewer labeled ``Analaysis Engines and Cas Consumers''. The workflow should look like the following:
We have now completed our example workflow. Addmittidely this is an incredibly simple workflow, but we hope this demonstrates how easy it is to combine components into workflows via U-Compare's drag and drop interface. To see how the workflow we constructed behaves, run it now.
Visualizer Components
After a while the session manager window should appear as it did with the NER Comparison Workflow in the previous section, but this time an ``Annotation Viewer'' window should also appear.
This additional window is opened due to the inclusion of a visualizer component in the workflow. The current version of U-Compare includes two such visualizer components, but more will be added in future.
The ``Annotation Viewer'' component which we included in this workflow is the first of these. It is used for graphically displaying the annotations of the CAS at the visualizers position in the workflow. This is of course just the same data we saw how to display in the ``Annotation Statistics'' tab of the session results window (which is in fact just an instantiation of this component). The reason for including the ``Annotation Viewer'' as an independent component is that it can be used to display annotations generated at any point in the workflow and not just those present when processing completes, which makes it a useful debugging tool.
The other visualizer component currently included with U-Compare is MoriV which is a tree structure visualizer capable of displaying and printing HPSG parse trees. The behavior of MoriV can best be seen by running the predefined ``U-Compare HPSG parse and visualize'' workflow.
Saving the Workflow
In case we want to use this workflow again in future, lets save it. To do this select the ``Save Workflow'' option from the workflow menu.
A dialog will appear asking you what name you want to use for the saved workflow, call it anything you like, but please do save it as we will be using it again later. Assuming the workflow was saved as ``Tutorial Workflow'' the workflow menu should have a new entry at the bottom as seen below:
The saved workflow can be loaded in exactly the same as loading the predefined workflows we saw earlier. The one difference is that user saved workflows appear with a red X icon to the left of their name, which when clicked gives an option of removing the saved workflow.
Section Summary
In this section we have seen how to create NLP workflows in U-Compare by simply dragging and dropping components into the order we want them to be executed. One of the main advantages of U-Compare is just how quick and easy it makes workflow creation. Of course this workflow is far too simple, so in the next section we will expand it into something more realistic.
Previous: Loading and Running Workflows Index: User Guide Next: Editing a workflow
